Hello, My name is Fred Trotter and I do some work with Free and Open Source Clinical information systems. The project that I have god-like control over is MirrorMed, which is a friendly-fork of ClearHealth. Currently, Genomics is a tool that is typically available to specialists and researchers, and unavailable to general practitioners. As we advance past decoding the human genome on the one hand, the majority of physicians in the US so not use computers to track patient health in their home practice. My hope is that FOSS clinical systems can help to make EHRs more pervasive and inexpensive. It is only after EHRs are common that we can truly usher in an era of genomic centered medicine. Rather than track the innovations that are available in proprietary systems, I would like to be innovating in advance of what the proprietary systems are capable of doing. As a result I would like to begin making my EHR, MirrorMed, Genomics aware. I would like to do this in a way that the other projects (FreeMED, OpenEMR, etc) will be able to borrow and improve on my work. For my first project I would like to find a way to cleverly use my EHR as a research tool. I am taking a class from Dr. Seungchan Kim over at ASU, and we have been working with analyzing gene expression data. As my final project I have elected to mold MirrorMed into a tool for studying gene expression data. MirrorMed can easily track clincal factors such as gender, smoking habits, age, race/ethnicity, etc etc. What I would like to do is be able to use a MirrorMed installation or a cluster of MirrorMed installations to provide good comparison gene expression data, in order to give feature analysis tools the ability to factor out clinical factors. I will be building a small prototype for this that I am already working on, but I wanted to see if your insights could guide that project into becoming long-term useful. Suppose I have gene expression data on 50 ovarian cancer patients, and I wanted to compare those to normal cells in 50 different normal patients. Now lets suppose that 35 of my patients are smokers, and 40 are over the age of 40. What I would like to do is be able to find a group of "matching" patients from within the MirrorMed system to select for comparison, in a HIPAA compliant manner. Given what MirrorMed is already capable of the easiest way to do that would be to use a web services model. I imagine a transaction of the services might look something like this. MirrorMed Research System-> Do you have any patients that match the following criteria -> Race = Chinese Smoking = 1/2 pack a day Between 40 and 45 years old Negative for Ovarian Cancer -> MirrorMed Clinical System MirrorMed Clinical System-> We have 30 such patients -> MirrorMed Research System MirrorMed Research System -> Please attempt to gather gene expressions data (Not sure how this would work) -> MirrorMed Clinical System MirrorMed Clinical System -> 13 patients have submitted to gene expression testing -> MirrorMed Research System MirrorMed Research System -> Request for de-identified clinical and gene expression data -> MirrorMed Clinical System MirrorMed Clincal System -> XML (MAGE? + CCR/HL7?) formatted data -> MirrorMed Research System I am just at the beginning of thinking about how to do this, but long term I would like this kind of a system to actually work. So I have several questions for your community. 1. Is this kind of thing done now, if so how? 2. What data formats are correct for transferring a combination of clinical and gene expression data? 3. Generally am I thinking about this the right way? 4. Is there a place other than this that is working on these kind of clinical/research hybrid solutions? 5. Is there an Internet resource that might inform my questions further. Google lead me to you. Generally its a problem of too much information, I need help sorting out what applies to me. I release all software, even mickey mouse code that is the results of school projects, under the GPL. -- Fred Trotter http://www.fredtrotter.com [[alternative HTML version deleted]]